Face Recognition Based On Automatic Calibration Feature Points

Face recognition aims at endowing computers with the ability to identify different human beings according to his/her face image. It has a wide range of potential application in the areas of public security, identification of certificate, entrance control and video surveillance. Although it is easy for most human observers to identify different human faces, automatically computerized face recognition is a very difficulty problem which remains largely unsolved.By referring to a great deal of the information available, various face recognition algorithms, such as based on geometric features, template matching and PCA, are researched and the advantages and disadvantages of them are analyzed in detail. The face recognition method based on geometric features is one of the early methods. Because of geometric features can not be extracted accurately so the result is not optimistic. A face recognition method based on automatic calibration feature points is presented here. Firstly, position eight key points of the human face accurately. Secondly, calculate eight important geometric features from feature points. Finally, use Continental distance to identify. Using eyes search algorithm to determine eye's region. A new Hybrid Projection Function is used to locate pupil position. Experiment proves that even if the pupil was not in the center of the eye, the pupil can still be accurate positioning. After positioning pupil, according to face structure characteristic determine regions of nose and mouse. Use distribution of gray and integral projection to locate feature points of nose and mouth separately.The experiment shows that this algorithm is superior to the traditional recognition algorithm, implementation of this method is fast, applicable to high real-time system.